Detecting Double H.264 Compression Based on Analyzing Prediction Residual Distribution

详细信息    查看全文

• 关键词：Double compression detection ; Prediction residual distribution ; Non ; aligned GOP structure ; First GOP estimation
• 刊名：Lecture Notes in Computer Science
• 出版年：2017
• 出版时间：2017
• 年：2017
• 卷：10082
• 期：1
• 页码：61-74
• 丛书名：Digital Forensics and Watermarking
• ISBN：978-3-319-53465-7
• 卷排序：10082

文摘

Detecting double video compression has become an important issue in video forensics. A novel double H.264 compression detection scheme based on Prediction Residual Distribution (PRED) analysis is proposed in the paper. The proposed scheme can be applied to detect double H.264 compression with non-aligned GOP structures. For each frame of a given video, the prediction residual is first calculated and the average value of the prediction residual in each non-overlapping 4 × 4 block is recorded to reduce the influence of the noise. Then the PRED feature is represented by the distribution of the average prediction residual in each frames. After that, the Jensen-Shannon Divergence (JSD) is introduced to measure the difference between the PRED features of adjacent two frames. Finally, a Periodic Analysis (PA) method is applied to the final feature sequence to detect double H.264 compression and to estimate the first GOP size. Fourteen public YUV sequences are adopted for evaluation. Experiments have demonstrated that the proposed scheme can achieve better performance than the state-of-the-art method investigated.